Study of the truncation strategy in the FPGA of a solar radio digital receiver
Fabao Yan,Yang Liu,Ke Xu,Ziqian Shang,Yuanyuan Zhang,Lei Zhang,Yanrui Su,Guang Lu,Zhao Wu,Yao Chen
DOI: https://doi.org/10.1093/pasj/psab010
2021-01-01
Publications of the Astronomical Society of Japan
Abstract:Computation resource is the limiting factor in higher operational accuracy of field-programmable gate arrays (FPGAs) in solar radio digital receivers. The data truncation strategy which determines the accuracy of data is then the essential technology in the design of a receiving system. Based on the solar radio spectrometer (dual channel, 14 bit, 1.25 gigasamples per second) at the Chashan Solar Radio Observatory (CSO), this paper presents a data truncation strategy which can realize real-time solar radio observation (35-40 GHz) with high time and frequency resolution as well as a large dynamic range, and at the same time saves the computation resource to a large extent. Simulations of truncations during signal processing are carried out in MATLAB, and the best truncation mechanisms are deduced for windowing and fast Fourier transform (FFT). Using the simulation results, the best truncation strategies have been implemented in the solar radio receiver at CSO with the result that the best truncation bits for the windowing operation are [27 : 14], with an error of 2.5 x 10(-4), and the best truncation bits for the FFT output are [20 : 5] with an error of 1.5 x 10(-3). Compared with the processing of full-precision data, occupation of the computation resources in the FPGA can be reduced significantly. For instance, the lookup table, lookup table RAM, flip flop, and digital signal processing slices are reduced by 7.36%, 14.65%, 8.38%, and 24.94%, respectively, which guarantees broad-band real-time solar radio observations (35-40 GHz).